Lecture 2 Flashcards
describe what our retinal image actually looks like
variable resolution (mostly poor)
wrong way up, and flipped side to side
if the eye only transmits borders why do I see objects as whole/complete?
some non-edge information is transmitted [to fill in the blanks], and the brain decoded the information from the eye to fill in the blanks (this happens even if they’re meant to be blanks)
what are receptive fields?
a region in the sensory periphery within which stimuli can influence the electrical activity of sensory cells
no retinal cell ‘sees’ all the image, each one only gets light from a small part of space
how do receptive fields work?
each photoreceptor receives a small part of information from the visual landscape
this portion of the retina that affects the signalling of that photoreceptor is called the cells’ receptive field
theres a part of the receptive field that turns cells on and turns cells off
how do receptive fields interact with photoreceptors?
photoreceptors will respond when light falls on their receptive field, whose output takes the form of action potentials in retinal ganglion cells whose axons form the optic nerve
what are retinal ganglion cells?
the output of the eye
rods and cones send their output to bipolar cells, and then bipolar cells send that output to the retinal ganglion cells, and it’s the axons from the retinal ganglion cells which make the optic nerve
what is the difference between rods and cones?
cones for colour, rods for low light vision
what are on-centre cells?
cells where responses are highest and most positive at the centre of the cell compared to the surrounding area (where there are negative responses)
how do on-centre cells work?
Since they constructed in such a way that when information, like light, activates the centre part of the receptive field, it will turn on, excite, or increase the firing rate of this retinal ganglion cell
whereas information which shines on the surround part of the receptive field will inhibit or turn off the retinal ganglion cell
what is the simultaneous contrast illusion “consequence of centre–surround antagonism”?
look at notes
give an example of retinal output signalling changes over space = edges
look at notes
edges are enhanced, areas of ‘no-change’ appear grey
another consequence of the centre-surround output from retinal ganglion cells is that they process edges
most are silent, but those receiving edge info are active (example of efficiency)
what is the Craik-O’Brien-Cornsweet Illusion?
look at notes
- The ganglion cells only respond to edges
- An image containing only edges (middle) passes straight through the retina more or less unchanged
- The normal image (right) gets converted to edges
- So the end result is that the output of the retina will ‘look’ much the same for the two images - just a bunch of edges
- So the rest of the brain interprets them as if they are the same - filling in the missing greys
- So they look more similar than they really are
describe Troxler fading and after-images
If you look at the blue-green cross at the top part of the scene for a few seconds and then after 10 secs look at the lower cross you should see the opposite colour underneath (white under black and vice versa)
They should start to fade and disappear because information about those colours is being inhibited, it will decrease
So Troxler fading and after images is a consequence of these compression mechanisms because energy is being saved
describe signalling changes over time and space
- Inhibition over time causes neurons to reduce their activity (save energy)
- Lack of crisp edges mean that spatial location is not signalled well either
- Reduced neural signal due to inhibition and lack of crisp edges allows filling-in of the grey regions by the cortex
- Thus, if a stimulus stays on for a long time it will appear to fade away - “Troxler fading”
- When the stimulus is removed the time lag on the inhibition produces the negative after image
what do the visual pathways look like?
look at notes
so information from the retina then crosses over at the optic chiasm before entering an important part of the visual system in the thalamus
describe how the lateral geniculate nucleus gateway to the cortex looks like?
look at notes for diagram
- K cells - colour (blue-yellow)
- P cells - slow response; colour (red-green) sensitive, fovea dominant, fine detail
- M cells - fast response; motion and depth sensitive, peripheral dominant, coarse detail
- Each layer of LGN contains a complete retinotopic representation of 1/2 visual field
- Cells have centre-surround receptive fields - just like Retinal Ganglion Cells
- No cell in each half of the LGN receives input from both eyes (no binocular cells)
But each LGN receives input from both eyes
bilateral structure (left is my left and right is my right)
describe the lateral geniculate nucleus
The LGN is constructed of several different layers
In between these layers are koniocellular cells which process information about blue and yellow
Parvocellular cells are like retinal ganglion cells so small cell bodies and respond more slowly to colour specifically red and green
They are also more closely concentrated in the fovea (the part of the retina where you’ve got high resolution vision) and they respond more to sort of finer detailed information
name the ventral (what) streams/pathways
V1, V2, V4, V8, LOC, OFA, FFA, PPA (important for recognising faces, places, objects)
from LGN to the what and where visual pathways
name the dorsal (where) streams/pathways
V3, MT/V5, V6, V7
(where objects are in space, motion, and depth)
from LGN to the what and where visual pathways
what is Prosopagnosia an example of?
Stroke in one of the what areas (FFA) leads to problems recognising faces (but not emotions)
what is Capgras syndrome an example of?
Stroke in one of the where areas (STS) leads to problems recognising emotions (but not faces)
how can the colour and lightness of objects be recovered?
not directly from the retinal image of a scene, but depend upon efficient neural processing by bottom-up (low-level spatial filters and feature detectors) and top-down mechanisms that incorporate contextual information and prior knowledge about the visual world
